An AI that compares genes across 1,000 species is speeding up the hunt for rare-disease causes
Hebrew University researchers published EvORanker, an algorithm that compares evolutionary gene patterns across more than 1,000 species, naming the correct disease gene as the top candidate in about 70 percent of cases and within the top five in 95 percent.
Artificial IntelligenceGNGV Editorial Team4 min read
Researchers at the Hebrew University of Jerusalem have built an artificial-intelligence algorithm, named EvORanker, designed to shorten one of medicine's most painful waits: the search for the genetic cause of a rare disease. Such a search can drag on for years and frequently ends with no answer, leaving families without a diagnosis or a path to treatment. The work, by Dr. Christina Canavati and Prof. Yuval Tabach of the Faculty of Medicine, was reported around the end of March 2026 and tied to a study in the journal Genetics in Medicine.
The tool's insight is evolutionary. EvORanker compares how genes have changed across more than 1,000 species, looking for hidden relationships between them, including links that current medical knowledge has never made. Genes that appear or disappear together over evolutionary time often share a function, so when a patient's DNA contains several candidate variants, the algorithm can use these deep patterns to rank which gene is most likely behind the disease.
In testing, the results were strong. EvORanker named the correct disease-causing gene as its top candidate in roughly 70 percent of cases and placed it within the top five candidates in about 95 percent, outperforming existing diagnostic tools especially for poorly understood genes. Prof. Tabach pointed to "thousands of cases like that around the world that fall through the cracks of current medicine," the kind of patients the team hopes to reach. The system is now available to researchers and clinicians, with applications expanding toward cancer research.
The caveats are clear. EvORanker ranks candidates and points doctors toward the most likely answer; it does not deliver a final diagnosis on its own, and broader clinical validation across diverse populations is still needed before routine use. It also depends on good genetic data, which is unevenly available worldwide. Still, for families trapped in a years-long diagnostic odyssey, an accurate tool that puts the right gene at the top of the list could turn an open-ended ordeal into a far quicker answer.
